Nutrition, Metabolism & Cardiovascular Diseases (2018) 28, 180e186

Available online at www.sciencedirect.com

Nutrition, Metabolism & Cardiovascular Diseases

journal homepage: www.elsevier.com/locate/nmcd

FreeStyle Libre and Dexcom G4 Platinum sensors: Accuracy

comparisons during two weeks of home use and use during
experimentally induced glucose excursions
F. Boscari a, S. Galasso a, A. Facchinetti b, M.C. Marescotti a, V. Vallone a, A.M.L. Amato a,
A. Avogaro a, D. Bruttomesso a,*
a
Division of Metabolic Diseases, Department of Medicine, University of Padova, Padova, Italy
b
Department of Informatics Engineering, University of Padova, Padova, Italy

Received 17 July 2017; received in revised form 12 October 2017; accepted 29 October 2017
Handling Editor: A. Giaccari
Available online 11 November 2017

KEYWORDS Abstract Background and aims: This study compared the accuracy of the FreeStyle Libre (Ab-
Type 1 diabetes; bott, Alameda, CA) and Dexcom G4 Platinum (DG4P, Dexcom, San Diego, CA) CGM sensors.
Continuous glucose Methods and results: Twenty-two adults with type 1 diabetes wore the two sensors simulta-
monitoring (CGM); neously for 2 weeks. Libre was used according to manufacturer-specified lifetime (MSL); DG4P
Flash glucose was used 7 days beyond MSL. At a clinical research center (CRC), subjects were randomized to
monitoring; receive the same breakfast with standard insulin bolus (standard) or a delayed and increased (de-
Accuracy; layed & increased) bolus to induce large glucose swings during weeks 1 and 2; venous glucose was
Hypoglycaemia; checked every 5e15 min for 6 h. Subjects performed
4 reference fingersticks/day at home. Ac-
Rate of change curacy was assessed by differences in mean absolute relative difference (%MARD) in glucose levels
compared with fingerstick test (home use) and YSI reference (CRC). During home-stay the Libre
MARD was 13.7  3.6% and the DG4P MARD 12.9  2.5% (difference not significant [NS]). With
both systems MARD increased during hypoglycaemia and decreased during hyperglycaemia,
without significant difference between sensors. In the euglycaemic range MARD was smaller with
DG4P [12.0  2.4% vs 14.0  3.6%, p Z 0.026]. MARD increased in both sensors following delayed
& increased vs. standard bolus (Libre: 14.9  5.5% vs. 10.9  4.1%, p Z 0.008; DG4P: 18.1  8.1% vs.
13.1  4.6%, p Z 0.026); between-sensor differences were not significant (p Z 0.062). Libre was
more accurate during moderate and rapid glucose changes.
Conclusions: DG4P and Libre performed similarly up to 7 days beyond DG4P MSL. Both sensors
performed less well during hypoglycaemia but Libre was more accurate during glucose swings.
Trial registration: The study was registered in ClinicalTrials.gov (NCT02734745) April 12, 2016.
ª 2017 Published by Elsevier B.V. on behalf of The Italian Society of Diabetology, the Italian Society
for the Study of Atherosclerosis, the Italian Society of Human Nutrition, and the Department of
Clinical Medicine and Surgery, Federico II University.

Introduction hypoglycaemia in individuals with type 1 diabetes (T1D)

[1e3]. Continuous glucose monitoring (CGM), offers real-
Good glycaemic control reduces the risk of micro- and time blood glucose values, alerts about impending hypo-
macro-vascular complications but increases the risk of glycaemia and hyperglycaemia, reduces HbA1c and has

* Corresponding author. Department of Medicine, University Hospital of Padova, Via Giustiniani 2, 35128 Padova, Italy.
E-mail address: daniela.bruttomesso@unipd.it (D. Bruttomesso).

https://doi.org/10.1016/j.numecd.2017.10.023
0939-4753/ª 2017 Published by Elsevier B.V. on behalf of The Italian Society of Diabetology, the Italian Society for the Study of Atherosclerosis, the Italian Society of Human
Nutrition, and the Department of Clinical Medicine and Surgery, Federico II University.
FreeStyle Libre and Dexcom G4 Platinum sensors
psychological benefits in T1D subjects treated with
continuous subcutaneous insulin infusion (CSII) and mul-
tiple daily insulin injection (MDI) [4e12]. Most glucose
sensors, however, require calibration, have short life time,
measure glucose in the interstitial fluid and cannot always
replace blood glucose self-monitoring (SMBG).
The FreeStyle Libre Flash glucose monitoring system
(FGM, Abbott Diabetes Care, Alameda, CA) is factory-
calibrated and has 14 days of wear lifetime but lacks
real-time alarms when limit glucose values are exceeded.
It is intended as a replacement for the SMBG, except dur-
ing times of rapid glucose changes, when hypoglycaemia
requires confirmation or when displayed glucose values
are not in accordance with symptoms.
A study evaluating the accuracy of Libre system at home
for 10e14 days [13] and another [14], comparing the ac-
curacy of Libre to that of two commercially available sen-
sors, found that Libre had comparable accuracy. To date, no
study has compared Libre’s performance with that of
comparators during rapid glucose swings or beyond sensor
certified wear lifetime.
The most commonly used CGM sensor in Italy is Dex-
com G4 Platinum (DG4P, Dexcom Inc., San Diego, CA),
which is provided by the health care system as adjunctive
to SMBG. The Libre system is purchased by patients or
provided by the health care system without adjunctive
SMBG for clinical decision-making.
The aim of this study was to compare the accuracy of
Libre vs. DG4P in adults with T1D as assessed by mean
absolute difference (% MARD) during glycaemic excursions
induced in at a clinical research center and during home
use.
Methods
Study design and participants
This was a prospective, single-centre, randomised, cross-
over study performed at the Clinical Research Center (CRC)
of Padova University between April and November
2016. The study was registered in ClinicalTrials.gov
(NCT02734745), approved by the institutional ethics review
board and done according the Declaration of Helsinki.
Written informed consent was obtained before study start.
Inclusion criteria were:
18 years, T1D for
1 year and
treatment with CSII or MDI.
Exclusion criteria were: pregnancy, lactation, medica-
tions impacting on glucose metabolism and inability to
comply with study procedures. Participants avoided drugs
impairing sensor-related measurement of glucose (i.e.
acetaminophen) during the study.
Study devices
The FreeStyle Libre System consists of a sensor and a
receiver/reader. The sensor is provided with a sterile
catheter (0.4 mm calliper), which is inserted 5 mm under
the skin and connected to a round disk (35 mm  5 mm)
applied on the back of the upper arm. The sensor measures
181
interstitial glucose levels every minute. By scanning the
reader over the sensor one obtains a real-time glycaemic
value, the glycaemic trend over the preceding 8 h, the
direction towards which glucose levels are moving and
glucose rate of change. The sensor lasts 14 days.
The Dexcom G4P consists of a small sensor inserted
under the skin which measures subcutaneous glucose
concentration every 5 min and sends readings to a
receiver. Twice-daily calibration with fingerstick blood
glucose is required. The manufacturer specified lifetime
(MSL) for the sensor is 7 days, however it is possible to
extend sensor use for a further week by reactivating it [15].
Procedures
Patients were trained on the use of the two devices at the
CRC. Sensors were simultaneously placed on the back of
one upper arm (Libre) and on the abdominal region (DG4P).
Home phase
During at-home CGM use, subjects were instructed to
perform SMBG at
4 times per day (before meals and at
bed-time) using the blood glucose (BG) meter built into
the hand-held Libre reader. Patients were instructed to
scan the sensor immediately after each BG test and to
based their insulin dosage adjustments on the BG reading.
Patients calibrated the DG4P according to manufactures’
specification against the BG measurements. After 7 days,
subjects began a second 7-day session with the original
DG4P sensor to assess accuracy beyond the manufacturer
specified lifetime. In case of sensor failure, loss of signal,
skin issues or accidental dislodgment, the sensor was
substituted. After 14 days, data from FreeStyle Libre and
DG4P were downloaded.
CRC studies
After calibration of the DG4P system, blood sampling
through a venous line started at 8:00 a.m. and glucose
concentration was measured on 25-mL-samples using YSI
2300 STAT PLUS glucose and lactate analyser (YSI Inc.,
Yellow Springs, OH).
At 8:15 a.m. subjects received a regular breakfast and
were randomized 1:1 to either their usual pre-meal insulin
dose or a delayed and increased bolus to induce significant
glycaemic excursions. At one visit, breakfast insulin dose
was calculated using carbohydrate-to-insulin ratio, adding
a correction bolus if fasting glucose was >5.6 mmol/L
(>100 mg/dL). Blood was sampled every 15 min between
8:00 to 11:00 a.m. and every 30 min during the next 3 h. At
the other visit, the insulin dose was doubled and admin-
istered 30 min after breakfast to induce an early post-meal
hyperglycaemia followed by a decrease in blood glucose.
Blood was sampled every 15 min between 8:00 a.m. to
9:00 a.m., every 10 min between 9:00 a.m. to 12:00 p.m.,
every 15 min between 12:00 p.m. to 1 p.m., and every
30 min between 1:00 p.m. to 2:00 p.m. Sensor readings
were obtained simultaneously to venous YSI samples and
capillary BG, which was tested hourly. If hypoglycaemia
was detected, subjects received rescue carbohydrates
182
when glucose was <3.0 mmol/L (<54 mg/dL) or at physi-
cian’s discretion and sampling was every 5 min in case of
hypoglycaemia. Patients left the CRC at 2:00 p.m.
Sensor accuracy assessments
The systems were tested over 14 days of continuous use at-
home and during two six-hour CRC admissions scheduled
at day 3e5 and 9e11 (Fig. S1). Accuracy was expressed as
mean absolute relative difference (%MARD). Plasma
glucose values were used as reference during the CRC
visits and SMBG values were used as reference during the
home phase. Sensor data were matched with venous (CRC
assessments) or capillary (home use assessment) BG
measurements. When CGM and reference values could not
be obtained at the same time, CGM values were linearly
interpolated to match reference values, provided CGM
values were obtained within 5 min of reference values.
MARD analyses were performed separately for break-
fast with normal bolus vs. breakfast with increased and
delayed bolus, and for week 1 vs. week 2 of sensor use.
MARD was calculated over the entire glycaemic range and
for glucose values: <3.9 mmol/L (70 mg/dL), and 3.9e10
mmol/L (70e180 mg/dL) and >10 mmol/L (>180 mg/dL).
Sensor accuracy was also assessed by calculating both the
percentage of data points in zones A and A þ B of the
Clarke Error Grid (CEG) [16] and the proportion of values
that fulfilled ISO 15197:2013 criteria (percentage of sensor
data within 15% of reference value for glucose concen-
trations
5.6 mmol/L (100 mg/dL), and within 0.8 mmol/
L (15 mg/dL)) of reference value for glucose concentrations
<5.6 mmol/l (100 mg/dL).
During CRC sessions, MARD was also calculated by
consolidating BG references into five groups based on the
rate of change (ROC), calculated as the first-order differ-
ence between the current and the previous sample divided
by the time distance between the two. The five ROC ranges
were: increasing glucose, >0.08 mmol/L/min (>1.5 mg/dL/
min) and >0.03 to 0.08 mmol/L/min (>0.5 to 1.5 mg/dL/
min); stable glucose, 0.03 to 0.03 mmol/L/min (0.5 to
0.5 mg/dL/min); and decreasing glucose, >0.03 to
0.08 mmol/L/min (>0.5 to 1.5 mg/dL/min), and >0.08
mmol/L/min (>1.5 mg/dL/min). Sensor accuracy based on
ROC values was measured considering all CRC sessions
(with and without induced glycaemic excursions), and
separating first from second week visits.
Because sensor accuracy was expected to be lower on
day 1 of sensor insertion [15,17] and during the second
week of monitoring (due to possible sensor performance
degradation), MARD analyses were also performed
comparing: day 1 vs. all the other days; days 1 and 8 vs. all
other days; week 1 vs. week 2; and combined days 1e10
vs. combined days 11e14. Data are presented as
mean  standard deviation.
Statistical analysis
Univariate analyses were performed for descriptive sta-
tistics. The t-test and the Wilcoxon signed-rank test were
F. Boscari et al.
used to compare normally and not normally distributed
variables. The means of more than two group were
compared with one-way analysis of variance (ANOVA). The
level of significance accepted was 0.05 using two-tailed
tests. Statistical analysis was performed using MATLAB,
Statistics Toolbox (Release 2016a, The MathWorks, Inc.,
Natick, Massachusetts, United States).
Results
Subject disposition
Twenty-four adults with type 1 diabetes were enrolled.
One subject was excluded from analysis as home data
could not be uploaded; one subject left the study for poor
device acceptance. Twenty-two subjects completed the
study: 13 female/9 male (10 CSII/12 MDI), mean age
36.3  12.9 years (mean  SD), diabetes duration
18.9  11.1 years, BMI 23.5  2.7 kg/m2, HbA1c 7.3  0.75%
(56.7  8.19 mmol/mol).
Ten subjects had breakfast with delayed and increased
bolus during the first week of sensor use, 12 during the
second week. The Libre device was scanned 14,4  6 time/
day.
Home phase
As shown in Fig. 1, there was no significant difference in
overall accuracy between Libre and DG4P during at-home
use. Both systems demonstrated similar performances in
the hypoglycaemic and hyperglycaemic ranges; however,
DG4P showed better accuracy in the euglycaemic range.
Both systems showed worse accuracy in the hypo-
glycaemic range.
Comparison of week 1 to week 2 showed similar ac-
curacy with both sensors during home use (Fig. 1). The
MARD for both systems worsened in week 2, but not
significantly (Libre: 13.0  3.7% week 1 vs. 14.6  4.9%
week 2, p Z 0.080; DG4P: 12.5  4.1% week 1 vs.
13.4  3.2% week 2, p Z 0.092) (Table S1).
As shown in Table S1, Libre’s accuracy worsened
significantly during days 11e14 compared with days 1e10
(12.6  6.0% on days 1e10 vs 15.0  8.2% on days 11e14,
p Z 0.006). DG4P had the worst accuracy on days 1 and 8
(14.5  7.1% vs. 11.7  7.4%, p Z 0.015), which was ex-
pected; CGM performance on day 1 of sensor wear is
known to be worse than performance on subsequent days.
CRC studies
Induced glycaemic excursions
MARD increased significantly in both the Libre and DG4P
systems following delayed and increased bolus at break-
fast; the betweenesensor difference was not significant
(Table 1). Libre showed a lower MARD vs. DG4P during
hyperglycaemia, whether induced by a breakfast with
delayed and increased bolus or following a breakfast with
correctly administered insulin bolus (Table 1). Overall
MARD values (with/without induced glycaemic
FreeStyle Libre and Dexcom G4 Platinum sensors 183

Figure 1 Top: Mean absolute relative difference (MARD) per day  95% confidence interval for Libre or DG4P. Bottom: Mean absolute relative
difference (MARD) between Libre or DG4P glucose readings and capillary glucose reference concentration at home.

Table 1 Mean absolute relative difference (%MARD) between Libre and DG4 P glucose readings and reference glucose concentration in venous
blood in subjects with type 1 diabetes receiving a breakfast with standard or delayed and increased insulin bolus.

Glucose profile Breakfast w/standard insulin bolus Breakfast w/delayed and increased insulin bolus
Libre DG4P Data pairs P value Libre DG4P Data pairs P value
Overall 10.9  4.1 13.1  4.6 424 0.055 14.9  5.5 18.1  8.1 710 0.062
Hypoglycaemia 10.8  6.9 12.9  11.9 9 0.824 21.7  14.4 27.3  22.9 87 0.148
<3.9 mmol/L (<70 mg/dL)
Euglycaemia 13.3  4.9 13.7  6.7 245 0.894 17.3  6.9 18.5  7.2 362 0.421
3.9e10 mmol/L (70e180 mg/dL)
Hyperglycaemia 7.8  4.5 11.2  5.1 170 0.010 10.2  4.9 14.5  6.1 261 0.031
>10 mmol/L (>180 mg/dL)
184
excursions) were similar during week 1; however, Libre
MARD was significantly lower during week 2 (Table S2).
Libre MARD values were also significantly lower in the
hyperglycaemic range during weeks 2 (Table S2).
Clark error grid analysis and fulfilment of ISO
15197:2013 criteria
There was no difference in the systems’ clinical perfor-
mance, most values fell within the clinically acceptable
error zones (A þ B) (Fig. 2, Table S3).
A notable proportion of glucose values from Libre and
DG4P fulfilled ISO 15197:2013 accuracy criteria: 70.2% and
73.5%, respectively, during home phase; 74.1% (Libre) and
65.0% (DG4P) during CRC assessment with normal insulin
bolus and by 69.5% (Libre) and 57.0% (DG4P) during CRC
assessment with increased and delayed insulin bolus
(Table S3).
Rates of glucose change
Analysis of accuracy during glucose swings was possible
only where YSI values were available. No significant dif-
ference in sensor accuracy was observed when glucose
was stable 0.03 to 0.03 mmol/L (0.5 to 0.5 mg/dL/min),
while Libre was significantly more accurate during periods
of moderate and rapid glucose changes (Table 2). Accuracy
among both sensors within the DG4P MSL (7 days), was
similar except during glucose changes at >0.08 mmol/L/
min (>1.5 mg/dL/min) (Table 2).
Sensors lifetime
Mean sensor lifetime was 13.45 days (1.3) for DG4P and
13.5 (1.2) for Libre. Eighteen (86.4%) Libre sensors func-
tioned for the full 14 days; 2 sensors were removed at day
10 and 1 at day 13 for adhesive issues and 1 sensor was
replaced at day 3 due to accidental dislodgment. Sixteen
(63.3%) DG4P functioned for the full 2 weeks, 1 sensor was
removed at day 9 and 4 at day 13 for adhesive problems,
Figure 2 Clarke Error Grid Analysis. Left: Libre (grey dots) and DG4P (black dots) vs. capillary measurements during 14 days at home. Right: Libre
(grey dots) and DG4P (black dots) vs. venous measurements after delayed and increased insulin administration.
F. Boscari et al.
and 1 sensor was interrupted at day 12 for acetaminophen
assumption by the subject.
Safety and adverse events
No sensor failure requiring removal or infection at the
insertion site were reported.
Discussion
This study compared the Libre and DG4P continuous
glucose monitoring systems during at-home use and dur-
ing induced glycaemic excursions. It is important to note
that the systems compared are not strictly equivalent,
since Libre data can be used in place of SMBG for insulin
adjustments, while DG4P is considered adjunctive to
SMBG. Furthermore, MSL is 14 days for Libre and only 7
days for DG4P. Despite these differences, we felt it was
important to compare the two systems because they are
widely used in Italy, and because many patients wear
DG4P beyond MSL.
In designing the study protocol, we relied on evidence
showing that DG4P accuracy remains unchanged 7 days
beyond MSL [15]. As shown in our study, we found that the
Libre and Dexcom G4 sensors functioned with similar ac-
curacy during home use for 2 weeks. Surprisingly, DG4P
accuracy was even better in the euglycaemic range, which
is likely due to the 2 daily calibrations recommended by
the manufacturer.
We also found that Libre’s accuracy decreased between
days 11 and 14, which is an important consideration for
patients who base insulin dosing on sensor readings only.
The increase in MARD and variability of DG4P on day 8 was
due to sensor recalibration, which is intrinsic to the Dex-
com algorithm [15].
With both systems, no difference in accuracy was found
comparing week 1 vs. week 2 (in CRC and at home). As
expected, the accuracy of both systems worsened
FreeStyle Libre and Dexcom G4 Platinum sensors 185

significantly during rapid glucose excursions due to the

P value

<0.001

<0.001
time delay between plasma and interstitial glucose [18].

0.019

0.039

0.248
Additionally, both sensors deviated approximately 20%
from reference value in the hypoglycaemic range during
Table 2 Mean absolute relative difference (%MARD) between Libre or DG4P readings and reference venous blood glucose according to different rates of change of glucose concentration. home use, indicating that low glucose readings should

Data pairs
routinely be confirmed by SMBG.
In studies done at the CRC during the first week Libre

144

157
71
55

66
and DG4P had similar accuracy across all glucose values
but Libre had a lower MARD when glucose decreased by

25.5  12.2
14.6  12.1

14.6  10.9

14.9  12.7
13.9  18.7
>0.08 mmol/L/min (>1.5 mg/dL/min). During the second
MARD (%)-YSI sessions

week Libre was more accurate than DG4P in the hyper-

DG4P

glycaemic range and when blood glucose levels increased

>0.03e0.08 mmol/L/min (>0.5 to 1.5 mg/dl/min). These
results are similar to a recent study of Aberer and col-
17.8  14.0

12.4  13.6
8e14 days

10.6  9.0

12.6  8.5

10.8  9.0
leagues, comparing Libre with DG4P and Medtronic Enlite
over 12 h in an inpatient setting under real-life conditions
Libre

(e.g., meals, exercise, hypoglycaemia and hyperglycaemia)

[14]. However, the effect of changes in glucose levels on
accuracy was not considered in the study.
P value

0.362

0.702

0.684
0.049

We found that Libre and DG4P had similar accuracy if

0.07

used within MSL and with glucose rates of change

<0.08 mmol/L/min (<1.5 mg/dL/min). For greater rates of
pairs
Data

change Libre was more accurate. It can be concluded that

177

180
106
99
72

in patients with good glycaemic control, using Libre or

DG4P would have little effect on dosing accuracy. How-
20.7  14.8
15.8  13.1

13.0  15.3

13.2  14.5
18.3  19.4

ever, Libre may be preferable for patients with high

MARD (%)-YSI sessions

glucose variability. We found no differences in sensor

DG4P

lifetime, nor in the need to replace sensors due to loss of

adhesion, accidental displacement or skin irritation.
The strength of this study is that it was performed in-
16.8  12.0
13.8  12.0

13.5  12.6

12.7  10.6
13.7  13.6

dependent from sensor manufacturers. Limitations may

1e7 days

include the fact that home glucose values were measured

Libre

by subjects themselves and that the DG4P has not been

approved by FDA for clinical-decision-making. Another
possible limitation is that home glucose values were
P value

<0.001

<0.001
<0.001
0.010

0.083

collected using the BG meter embedded in the Libre sys-

tem. Although Libre is factory calibrated and does not use
capillary glucose values, the factory calibration could have
pairs
Data

been designed and/or optimized to better match the blood

170
127

321

337
172

glucose values collected with the Libre meter. However,

the results of the Libre’s accuracy at home vs CRC, where
23.0  14.1
15.2  12.7

14.0  13.7

14.3  13.6
16.9  18.9
MARD (%)-All YSI sessions

the BG references have been collected with an indepen-

dent system (YSI), do not evidence any advantage for the
DG4P

Libre. It should be noted that the DG4P sensor used in the

study utilized an older algorithm, which has since been
updated (505 software). Dexcom has also launched a new
17.0  14.7
12.4  11.0

12.9  11.0

13.2  13.3
1e14 days

11.8  9.9

sensor (Dexcom G5 Mobile), which is the first CGM device

approved by the U.S. Food and Drug Administration (FDA)
Libre

for non-adjunctive use. It would be interesting to compare

Libre with the Dexcom G5 Mobile system.
>0.03 to 0.08 (>0.5 to 1.5)

0.03 to 0.03 (0.5 to 0.5)

>0.03 to 0.08 (>0.5 to 1.5)

mmol/L/min (mg/dL/min)

Conclusion

We demonstrated that DG4P is as accurate as Libre over 2

Rate of change,

weeks irrespective of MSL. Libre and DG4P have similar

>0.08 (>1.5)

>0.08 (>1.5)
Decreasing
Increasing

accuracy during rapid glucose swings if used within MSL

and when glucose changes by <0.08 mmol/L/min
Stable

(<1.5 mg/dL/min). Above this rate of change, Libre is more

accurate.
186
Funding
This research did not receive any specific grant from
funding agencies in the public, commercial, or not-for-
profit sectors.
Author contributions
D.B. and F.B conceived the study, had full data access,
drafted the manuscript and are responsible of the integrity
of the data. S.G., A.F., and A.A. critically reviewed the
manuscript. A.F. analyzed the data. M.C.M. analyzed blood
sample. V.V. and A.M.L.A. collected the data.
Conflicts of interest
Authors declare no competing interests.
Acknowledgements
The authors thank subjects who participated to the study.
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.numecd.2017.10.023.
References
[1] Diabetes Control and Complications Trial Study Group. The effect
of intensive treatment of diabetes on the development and pro-
gression of long-term complications in insulin-dependent dia-
betes mellitus. N Engl J Med 1993;329:977e86.
[2] DCCT Research Group. Epidemiology of severe hypoglycemia in
the diabetes control and complications trial. Am J Med 1991;90:
450e9.
[3] Davis S, Alonso MD. Hypoglycemia as a barrier to glycemic control.
J Diabetes Complicat 2004;18(1):60e8.
[4] Deiss D, Bolinder J, Riveline JP, Battelino T, Bosi E, Tubiana-Rufi N,
et al. Improved glycemic control in poorly controlled patients with
type 1 diabetes using real-time continuous glucose monitoring.
Diabetes Care 2006;29(12):2730e2.
[5] Hermanides J, Nørgaard K, Bruttomesso D, Mathieu C, Frid A,
Dayan CM, et al. Sensor augmented pump therapy lowers HbA1c
in suboptimally controlled Type 1 diabetes; a randomized
controlled trial. Diabet Med 2011;28:1158e67.
F. Boscari et al.
[6] Hommel E, Olsen B, Battelino T, Conget I, Schütz-Fuhrmann I,
Hoogma R, et al. Impact of continuous glucose monitoring on
quality of life, treatment satisfaction and use of medical care re-
sources: analyses from the SWITCH study. Acta Diabetol 2014;51:
845e51.
[7] Pickup JC, Freeman SC, Sutton AJ. Glycaemic control in type 1
diabetes during real time continuous glucose monitoring
compared with self monitoring of blood glucose: meta-analysis of
randomised controlled trials using individual patient data. BMJ
2011;343. d3805.
[8] Heinemann L, DeVries JH. Evidence for continuous glucose
monitoring: sufficient for reimbursement? Diabet Med 2014;31:
122e5.
[9] Lind M, Polonsky W, Hirsch IB, Heise T, Bolinder J, Dahlqvist S,
et al. Continuous glucose monitoring vs conventional therapy for
glycemic control in adults with type 1 diabetes treated with
multiple daily insulin injections: the GOLD Randomized Clinical
Trial. JAMA 2017;317:379e87.
[10] Battelino T, Conget I, Olsen B, Schütz-Fuhrmann I, Hommel E,
Hoogma R, et al., The SWITCH Study Group. The use and efficacy of
continuous glucose monitoring in type 1 diabetes treated with
insulin pump therapy: a randomised controlled trial. Diabetologia
2012;55:3155e62.
[11] Beck RW, Riddlesworth T, Ruedy K, Ahmann A, Bergenstal R,
Haller S, et al. Effect of continuous glucose monitoring on glycemic
control in adults with type 1 diabetes using insulin injections: the
DIAMOND Randomized Clinical Trial. JAMA 2017;317:371e8.
[12] The Juvenile Diabetes Research Foundation Continuous Glucose
Monitoring Study Group. Continuous glucose monitoring and
intensive treatment of type 1 diabetes. N Engl J Med 2008;359:
1464e76.
[13] Ólafsdóttir AF, Attvall S, Sandgren U, Dahlqvist S, Pivodic A,
Skrtic S, et al. A clinical trial of the accuracy and treatment
experience of the Flash glucose monitor FreeStyle Libre in adults
with type 1 diabetes. Diabetes Technol Ther 2017;19(3):164e72.
[14] Aberer F, Hajnsek M, Rumpler M, Zenz S, Baumann PM, Elsayed H,
et al. Evaluation of subcutaneous glucose monitoring systems
under routine environmental conditions in patients with type 1
diabetes. Diabetes Obes Metab 2017 Feb 16. https:
//doi.org/10.1111/dom.12907 [Epub ahead of print] available from:
https://www.ncbi.nlm.nih.gov/pubmed/28205324.
[15] DeSalvo DJ, Ly TT, Wadwa RP, Messer L, Westfall E, Gopisetty D,
et al. Continuous Glucose Sensor survival and accuracy over 14
consecutive days. Diabetes Care 2016;39(8):e112e3.
[16] Wentholt IM, Hart AA, Hoekstra JB, Devries JH. How to assess and
compare the accuracy of continuous glucose monitors? Diabetes
Technol Ther 2008;10(2):57e68.
[17] Christiansen M, Bailey T, Watkins E, Liljenquist D, Price D,
Nakamura K, et al. A new-generation continuous glucose moni-
toring system: improved accuracy and reliability compared with a
previous-generation system. Diabetes Technol Ther 2013;15(10):
881e8.
[18] Cobelli C, Schiavon M, Dalla Man C, Basu A, Basu R. Interstitial fluid
glucose is not just a shifted-in-time but a distorted mirror of blood
glucose: insight from an in silico study. Diabetes Technol Ther
2016;18(8):505e11.